Malaria Data Modeling workshop

Introduction to uv, VSCode, Streamlit and Aider

📱 Follow Along on Your Device

Scan this code to view the slides on your phone or laptop:

What We’ll Cover Today

Installing Tools - uv and Aider setup
Project Initialization - Creating your workspace
Configuration - OpenRouter API and Aider settings
Installing Dependencies - Adding required libraries
The Development Workflow - Building with AI
Running the Application - Launching your dashboard

Part 1: Installation & Setup

Install uv

uv is an extremely fast Python package and project manager, written in Rust 🦀

Windows (using winget)

winget install --id astral-sh.uv

macOS & Linux (Ubuntu)

curl -LsSf https://astral.sh/uv/install.sh | sh

Check if installed correctly

uv --version

Install uv

uv is an extremely fast Python package and project manager, written in Rust 🦀

From Github releases

https://github.com/astral-sh/uv/releases

Why is uv So Fast?

Key Features

10-100x faster than pip
Unified tool - replaces pip, poetry, pyenv
Reliable - global caching & lockfiles
Cross-platform - Mac, Windows, Linux

Built with Rust 🦀

Compiled to machine code
Parallel processing
No GC pauses
Minimal memory

Install Aider Globally

Using uv

uv tool install aider-chat --python 3.12 --force

What is Aider?

AI pair programmer in your terminal
Direct file editing
Git integration
Whole codebase context
Works with any LLM (we’ll use Gemini)

Part 2: Project Initialization

Download from Github repo

Download from Github repo

Unzip the file and cd into it

# move into the directory
cd MalariaDataModelingWorkshop-main

Part 3: Configuration

OpenRouter

Why we are choosing it:

Unified Access: Connect to Deepseek, OpenAI, Google, etc via one key.
Privacy First: Settings for “Zero Data Retention.”
Future Proof: New models are available almost immediately.
Best Prices: easy access to low-cost open-source models.

A “Universal Adapter” for AI

OpenRouter

Get the API KEY

https://openrouter.ai/

OpenRouter

Get the API KEY

Create Aider Config File

check the directory for the file below and edit:

example.aider.conf.yml

# ~/.aider.conf.yml

# 1. Add your openRouter API key
# Get one from https://openrouter.ai/settings/keys
api-key:
 - openrouter=sk-...
 

# 2. Set your default model
# Assuming we are using Deepseek...
model: deepseek/deepseek-v3.2
# 3. Optional: Enable caching to save on costs
cache-prompts: true

change the name to .aider.conf.yml and save

Security: Update .gitignore

Always protect your API keys

# Add to .gitignore
.aider.conf.yml
.env
*.env

Part 4: Installing Dependencies

Install Required Libraries

Add all dependencies upfront

uv sync

Why Install Dependencies First?

Aider will use them during test generation
TDD (Test-Driven Development) requires dependencies to be available
Ensures tests can run immediately
Prevents import errors during development

Part 5: The Development Workflow

Launch Aider

Basic Command

aider --no-auto-commits

You’ll see:

Aider v0.x.x
Model: gemini/gemini-1.5-pro-latest
Git repo: .git
>

If the last launch fails…

…try the following

uv run --with aider-chat python -m aider --model openrouter/Deepseek/deepseek-v3.2 --api-key openrouter=YOUR_OPENROUTER_API_KEY

The Four-Step Process

Context & Requirements - Set up the “Expert” persona
Freeze the Spec - Lock requirements to prevent scope creep
Planning & TDD - Break work into testable tasks
The Loop - Iterate through tasks automatically

Step A: Context & Requirements

The “Expert” Persona Approach

Copy the prompt from the PROMPT.md file and paste

Step B: Freezing the Spec

Lock Requirements

Once the PRD is generated, prevent AI from changing it:

/read-only context/PRD.md

Why?

Prevents scope creep
Maintains focus
Ensures consistency

Step C: Planning & TDD

Create Task List

This was already accomplished from the instructions on the PROMPT.md. You can check the folder visually to confirm.

What Aider Creates

Example Task List

## Tasks

- [ ] 1. Set up project structure
- [ ] 2. Create file upload component
- [ ] 3. Implement CSV validation
- [ ] 4. Add EDA functions
- [ ] 5. Create visualization module
- [ ] 6. Build main dashboard layout
- [ ] 7. Write tests for each component

Step D: The Loop

The Simple Iteration Command

Please proceed to the next task.

That’s it!

Aider checks off completed tasks
Moves to the next item
Writes code + tests
Commits to git
Repeat until done

Essential Aider Commands

/add file.py - Add file to context
/read-only file.py - Prevent modifications
/drop file.py - Remove from context
/ls - List files in context
/undo - Undo last change
/diff - Show pending changes
/help - Show all commands
/exit - Exit aider

Part 6: Running the Application

Launch the App

Using uv run

uv run streamlit run app.py

No activation needed!

uv run automatically uses the virtual environment

Key Principles

Isolated environments

Use uv for reproducible dependencies
Fast, reliable, consistent

Resources

Documentation

uv: docs.astral.sh/uv
OpenRouter: https://openrouter.ai/
Aider: aider.chat/docs
Streamlit: docs.streamlit.io

What’s Next?

Set up your own dashboard project
Try the expert persona approach
Experiment with task decomposition
Share your results with the team
Build more AI-assisted projects!

Let’s Build Something Amazing! 🚀

Your Turn

Your next project

Create a new folder for your new project
From this project copy the pyproject.toml & .aider.conf.yml files

Let’s Build Something Amazing! {transition=“fade”}🚀

Your Turn

Your next project

Paste these files into your new project
Open the terminal from this new project
Install dependencies with uv sync
Follow the 4-step workflow
Create your own PROMPT.md. Use ChatGPT to help you.
Deploy your fresh dashboard!

Summary: Modern Python Development

⚡ uv - 10-100x faster package management
🤖 Aider - AI pair programmer in your terminal
🎯 OpenRouter - The Unified Interface for LLMs
📊 Streamlit - Rapid dashboard development
🔄 Iterative workflow - Expert persona + task loops

Questions?

Thank You! 🎉

Happy Vibe Coding! 💻✨

Get the entire files here.